The ever-increasing importance of electronics in automobiles brings with it a growing challenge and need for low-cost, reliable electronic systems and subsystems such as sensors and actuators. These systems and subsystems are not isolated, but must communicate with each other.

Historically, automotive electronics have been built up using discrete, smaller integrated circuits. They relied on proprietary, dedicated wire communication schemes, at least for many sensor systems, and directly wired power outputs to the actuators. This led to large printed-circuit boards (PCBs), large engine-control unit (ECU) housing sizes, and excessive wiring. Wiring brings with it other problems since it consumes space, adds weight and expense, is subject to the vehicle's electromagnetic noise, and can be difficult to maintain.

Fortunately, advances in vehicle-networking standards and mixed-signal semiconductor processes are addressing these issues and introducing new possibilities to distribute intelligent systems throughout a vehicle. The trend in vehicle-networking standardization includes the wide adoption of Controller Area Network (CAN) and the Local Interconnect Network (LIN) architecture, now in version 2.0.

These network standards are providing a balance between performance and cost optimization across automotive systems. CAN provides a high-speed network for chassis, power-train and body-backbone communications, while LIN answers the need for a simple network for sensor and actuator subsystems that reduces cost and improves robustness through standardization. The wide use of CAN and the availability of LIN coincides with advances in mixed-signal semiconductor-process technologies that can bring together all the functionality needed for smaller automotive systems onto a single IC, or a few ICs for more advanced systems.

Together, the vehicle networking standards and advanced mixed-signal processes provide an opportunity for automotive manufacturers to introduce affordable new electronic systems as well as cost-reduce existing systems. They also improve maintenance and reliability while providing advanced convenience and safety features to a car's occupants.

Vehicle Networking
The CAN standard is well established and has been refined as the defacto standard for communication between systems and ECUs. CAN is a two-wire, high-speed differential bus with a specified data rate of up to one megabits per second (Mbps). However, it is typically used at a data rate between 125 kilobits per second (kbps) and 500 kbps. CAN uses a special differential driver that provides either a positive differential voltage or presents high impedance on the bus.

This physical layer feature allows lossless arbitration known as Carrier Sense, Multiple Access/Collision Detection + Collision Resolution (CSMA/CD+CR) defined in the CAN protocol. Any node can access the bus after the previous message has been sent. If two nodes try to send a message at the same time, the node sending the highest priority message will win the arbitration and continue transmitting its message without restarting or losing any of the data that was already sent.

Implementation of a CAN network in an ECU typically includes a microprocessor (microcontroller or digital signal processor) with an integrated CAN protocol controller, along with a CAN transceiver and other supporting circuitry. Normally the complexity of these systems is too high to integrate into a monolithic IC. However, advances in mixed-signal processes and integration capability are leading to many system base ICs and mixed-signal ASICs that may allow for a two-chip solution (MCU and mixed-signal device), or at least reduce the number of ICs needed in the ECU.

While LIN was originally targeted for the vehicle's body electronics, it is proving its value in new ways with many implementations outside of body electronics. Among the automotive-electronic bus standards available, LIN provides the best solution for the communication needs of most sensors and actuators which are normally dedicated to a single system. They can be viewed as subsystems and are well served by LIN, which has been defined to fill a sub-network role in the vehicle. The maximum LIN specified data rate of twenty kbps is sufficient for most sensors and actuators. LIN is a time-triggered, master-slave network, eliminating the need for arbitration among simultaneously reporting devices. It is implemented using a single wire, which reduces wiring and harness requirements and thus helps save weight, space, and cost.

Defined specifically for low-cost implementation of vehicle sub-network applications by the LIN Consortium, the standard aligns well to the integration capabilities of today's mixed-signal semiconductor processes. The LIN protocol achieves significant cost reduction since it is fairly simple and operates via an asynchronous serial interface (UART/SCI), and the slave nodes are self-synchronizing and can use an on-chip RC oscillator instead of crystals or ceramic resonators. As a result, silicon implementation is inexpensive, making LIN very suitable for the mixed-signal process technologies typically used to manufacture signal-conditioning and output ICs for automotive subsystems.

The LIN master node is normally a bridge node of the LIN sub-network to a CAN network, and each vehicle will typically have several LIN sub-networks. The master LIN node has higher complexity and control, while the slave LIN nodes are typically simpler, enabling their integration in single IC subsystems.

Both LIN and CAN have replaced many communication implementations that existed, such as previous generation networking schemes, pulse-width modulation (PWM), and variable pulse-width (VPW) architectures. These implementations are based on various physical-layer designs. Most of the non-networking schemes were unidirectional and required at least one dedicated wire per signal. As a result, these architectures offer limited or no possibility for two-way communication and diagnostics. Also, because these solutions are often proprietary, they did not take advantage of the economies of scale and design reuse achieved through implementation of an open standard.

Advanced mixed-signal processes The widespread adoption of CAN and LIN standards is an important development for automotive electronics, and becomes much more significant in conjunction with recent advances in mixed-signal semiconductor processes. Today, IC manufacturers who leverage expertise in both high-speed CMOS digital and advanced mixed-signal/analog processes are implementing levels of system integration that were not imagined just a few years ago.

Representative types of advanced mixed-signal processes that can be used for automotive applications are linear Bi-CMOS (LBC or BCD), high voltage CMOS, and Silicon-on-Insulator (SoI). Many of these processes will allow for a monolithic system-on-chip (SOC) implementation or implementation in a few ICs of the entire automotive electronics, including physical layer interface, power, high-voltage, digital logic, memory and precision analog functions.

For systems where on-chip intelligence is needed, advanced mixed-signal processes enable the integration of a reasonable level of digital logic, hardwired digital processing, network protocol engines, and small microprocessors. For instance, mixed-signal designs might include logic (state machine and protocol engine) that can respond to standard networking commands via LIN, to control and report the status of a sensor or actuator. System-on-a-chip integration such as this will be invaluable for an application such as a "one-touch" window lifter, which needs to run an algorithm that keeps the rising glass from pinching fingers, reports problems within the system, and can provide diagnostics to technicians. Complex applications requiring higher-speed communications can use the same semiconductor capabilities that make it possible to integrate LIN communication to integrate CAN capabilities on mixed-signal ICs.

Automotive electronic subsystem examples As an example of an electronic subsystem based on LIN, look at Texas Instrument's TPIC1021 LIN-2.0 transceiver and catalog (standard) voltage regulators, based on their LBC processes. From this starting point, you can integrate the additional components needed in an LBC process, to build a system on the chip that will interface to the vehicle's electrical network as well as a LIN network, while providing all necessary functions in a robust, yet affordable IC. Typical system-on-chip functions include an automotive voltage regulator matched to the system requirements, a networking transceiver, analog filtering of sensor inputs, power outputs, an analog-to-digital converter (ADC), digital filtering and control, and a network protocol controller. Examples of a fully integrated sensor subsystem based on LBC are shown in Figure 1.

The high-level of integration and circuit protection make the device well-suited to the space and cost constraints of harsh automotive environments.

Figure 1: LBC Sensor Subsystem System on Chip (SOC)
(Click to Enlarge Image)

Figure 2 shows an example of a system base chip developed in an LBC process that reduced the number of semiconductor devices necessary in previous generation actuator subsystem while adding functionality.

Figure 2: LBC System Base Chip (SBC)
(Click to Enlarge Image)

Such a subsystem could co-exist on a LIN sub-network and work with sensor subsystems as illustrated in Figure 1. This system base chip IC integrates a voltage regulator from vehicle battery source, a voltage supervisor and reset circuit, high-voltage interface to the user switches, high-side driver, two low-side drivers for relays that may control a motor or high current load, op amp for control loop feedback, protection circuitry, and LIN 2.0 compliant transceiver. This device interfaces directly to the system's microcontroller which provides the control algorithms such as anti-pinch supervision in window lifters. It also provides the LIN protocol handing for this subsystem.

For other applications, the same mixed-signal process technology is capable of integrating the blocks in these two devices along with other functions, including low-dropout and switching voltage regulators for single and multiple rails, different configurations of high- and low-side drivers, various op amps, digital logic, and vehicle networking interfaces such as LIN and CAN.

Possible power-output modules include H-bridge type of intelligent drivers, for single-phase DC brush motors, and three-phase DC brush motors and relay drivers. These ICs are used in a wide range of automotive systems including chassis, power train, power seats and mirrors, door locks, windshield wipers and defrosters, window and antenna lifters, heating, ventilation and air-conditioning (HVAC), and a variety of other electronic systems for user comfort and security.

Automotive electronic system benefits
Adopting standardized vehicle-networking architectures, and using more highly integrated mixed-signal ICs, brings several advantages at the system level. The first is an improvement in a system's robustness and diagnostics. By adopting standardized networks for two-way communication, diagnostic and failure information can be obtained when there are issues with the system. Removing proprietary interfaces allows the development of systems and software that use a common communication scheme based on a known, reliable standard, and also provide the opportunity for greater reuse in the future.

The second is reduced wiring requirements. Through the use of standardized vehicle-networking architectures, it is possible to build a feature- and diagnostic-rich system that requires only three wires (LIN: battery, ground and LIN) or four wires (CAN: battery, ground, CANH, and CANL). Reduced wiring requirements have less cost, less weight, easier installation at the factory, and a reduction in potential sources of failure.

Integration also leads to other advantages and savings. PCBs and housings can be smaller, allowing for improved and more flexible placement in the vehicle, with less concern about where and how to run the wires. Through the use of fewer components, there are fewer items to keep in inventory, qualify, and monitor. Some of these factors also lead to a reduction in weight and space consumption, factors that are always at a premium in vehicle design.

This advance is another step in increasing the intelligence and capabilities of automotive systems. The next generation of mixed-signal automotive ICs will integrate even more performance and processing power. They will provide programmable features and added flexibility that will be used to address the automotive electronic system needs of tomorrow. As these systems become more advanced, the possibilities are limited only by the applications that vehicle designers can imagine and the end customer is willing to purchase.

About the author

As Systems Architect, Mixed Signal Automotive Group, Texas Instruments, Scott Monroe has worked in the automotive semiconductor market for more than a decade. A graduate of Rose-Hulman Institute of Technology, he has delivered numerous papers at automotive conferences and written several contributed articles ranging from microcontrollers and software in safety systems through automotive networking physical layers. Scott can be reached at s-monroe1@ti.com.